Agricultural Economics presented on for the (Name of student)

advertisement
AN ABSTRACT OF THE THESIS OF
for the
SHENG HUI LIAO
DOCTOR OF PHILOSOPHY
(Degree)
(Name of student)
in
Agricultural Economics
presented on
(Major)
Title:
May 16, 1973
(Date)
AN ANALYSIS OF THE DIFFERENCES IN THE MARGINAL
PROPENSITY TO CONSUME OF FARM. RURAL NONFARM
AND URBAN FAMILIES IN THE
Abstract approved:
ITED STATES. 1961
Redacted for Privacy
John A. Edwards
The study was conducted to obtain information concerning the
factors affecting the differences in the marginal propensities to consume of farm and nonfarm families observed by Friedman and Lee
and Phillips. As a first step a theoretical framework was developed
which indicated the factors affecting the marginal propensity to con-
sume. From this framework, two hypotheses were formulated for
explaining observed differences in the marginal propensities to
consume between farm and nonfarm families: (1) Farm and nonfarm consuming units with homogeneous socioeconomic character-
istics have the same marginal propensity to consume, and (2) distribution of consuming units by socioeconomic characteristics in the
farm sample is different from that of consuming units in the nonfarm sample.
Next a model and statistical procedure for testing the hypotheses were developed. The dummy variables technique was used as
a means of quantifying the socioeconomic variables and thus of
measuring their effect on the marginal propensity to consume.
Finally, the model was fitted to data obtained from the Bureau of
Labor Statistics and the U. S. Department of Agriculture--Survey
of Consumer Expenditure, 1960-61--using least squares procedures.
The empirical results indicated that there were no significant
differences in the marginal propensities to consume between farm
and nonfarm families with homogenous socioeconomic character-
istics for the majority of the 17 tested groups, However, those
groups in the Northcentral and Western regions were the exception.
In these two regions, farm families had a lower marginal propensity
to consume than did nonfarm families.
This could be due to
differences in the prices paid by farm and nonfarm families in these
two regions.
The empirical results also indicated that there were significant
differences in the distribution of families by socioeconomic char-
acteristics between the farm and nonfarm samples. Thus, the
observed differences in the marginal propensities to consume between
farm and nonfarm families could be due to the differences in the
distributions of family types in the two samples.
An Analysis of Differences in the Marginal Propensity
to Consume of Farm, Rural Nonfarm, and Urban
Families in the United States, 1961
by
Sheng Hui Liao
A THESIS
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Doctor of Philosophy
June 1974
APPROVED:
Redacted for Privacy
Professor of Agricultural Economics
Iin charge of major
Redacted for Privacy
Head of Department of Agricultural Economics
Redacted for Privacy
Dean of"Graduate School
Date thesis is presented
May 16, 1973
Typed by Velda D. Mullins for
Sheng Hui Liao
ACKNOWLEDGMENT
The author owes a special debt of gratitude to
Dr. John A. Edwards, major professor, for his constructive suggestions at all stages of this study and
guidance throughout the authorTs graduate program at
Oregon State University.
He is grateful to Drs. Timothy M. Hammonds, Joe
B. Stevens, Bruce R. Rettig, Richard S. Johnston, William
G. Brown, and Mr. Ancel D. Haroldsen for their helpful
comments and assistance on the dissertation.
He also wishes to express his sincere appreciation
to the Department of Agricultural Economics, Oregon State
University, for providing financial assistance during the
entire period of his graduate study and also to the Computer
Center, Oregon State University, which provided a grant in
support of this study.
The authors greatest debt is to his wife for her
patience and her sacrifices throughout his graduate work.
TABLE OF CONTENTS
Page
Chapter
I
II
INTRODUCTION
1
THEORETICAL FRAMEWORK
6
Consumer Choice and the Marginal Propensity
to Consume
The Formulation of Hypotheses
III
IV
THE MODEL AND STATISTICAL PROCEDURE
12
The Model
12
The Statistical Procedure
13
EMPIRICAL RESULTS
The Data
The Definitions
The Marginal Propensity to Consume Out of
Disposable Money Income
The Marginal Propensity to Consume Out of
Total Disposable Income
V
6
10
CONCLUSIONS AND IMPLICATIONS
Validity of the Hypotheses
Comparison with Previous Studies
Methodological Conclusions
Policy Implications
Implications for Future Research
21
21
22
26
56
65
65
68
71
73
75
BIBLIOGRAPHY
76
APPENDIX A The Equality of Regression Coefficients of
the Dummy Variables Regression Model and
the Ordinary Regression Model
79
APPENDIX B The Distribution of Families in the Sample
by Socioeconomic Characteristics
85
LIST OF TABLES
Page
Table
1
2
Regression coefficients and related statistics
of urbanization equation in the U.S., 1961
Differences in the marginal propensities to consume
and their t-values for farms, rural, nonfarm, and
urban families in the U. S.,
3
4
5
6
7
8
9
10
11
28
1961
Regression coefficients and related statistics of
regionality equation in the U. S., 1961
30
31
Differences in the marginal propensities to consume
and their t-values for four regions in the U. S.,
1961
32
Regression coefficients and related statistics of
education equation in the U.S., 1961
33
Differences in marginal propensities to consume
and their t-values for four educational levels of
family head in the U.S., 1961
34
Regression coefficients and related statistics of
ages equation in the U.S., 1961
35
Differences in the marginal propensities to consume
and their t-values for six age groups in the U. S.,
1961
36
Regression coefficients and related statistics of
family size equation in the U.S., 1961
38
Differences in the marginal propensities to consume
for seven family size groups in the U.S., 1961
40
Regression coefficients and related statistics of
occupation equation in the U.S., 1961
41
Table
12
13
14
15
16
17
18
19
20
21
22
Page
Differences in the marginal propensities to
consume for seven occupational groups in the
U.S., 1961
44
Multiple regression coefficients and related
statistics in farm and nonfarm families, U. S.,
1961
47
Socioeconomic characteristics of the family in
each group
50
Distribution of families in the farm and the nonfarm samples by characteristics of the family
51
Consumption functions and related statistics for
selected homogeneous groups of farm and nonfarm
families in the U.S., 1961
53
Regression coefficients and related statistics of
urbanization equation and three urbanization
families in the U.S., 1961
58
Differences in the marginal propensities to
consume out of total disposable income and their
t-values for farm, rural nonfarm and urban
families in the U.S., 1961
59
Multiple regression coefficients and related
statistics to farm and nonfarm families, U.S., 1961
60
Consumption functions and related statistics for
selected homogeneous groups of farm and nonfarm
families in the U.S., 1961
62
Distribution of families in the farm and the nonfarm samples based on socioeconomic characteristics
67
Comparison of current study and Friedman's study:
the marginal propensities to consume of farm, rural
nonfarm, and urban families in the U. S.
69
AN ANALYSIS OF DIFFERENCES IN THE MARGINAL
PROPENSITY TO CONSUME OF FARM, RURAL
NONFARM, AND URBAN FAMILIES IN THE
UNITED STATES, 1961
I INTRODUCTION
The marginal propensity to consume is defined as the change
in consumption expenditure in response to a change in income. In
simple macro-models the marginal propensity to consume is as-
sumed to be essentially the same for all consumers. However, empirical studies by Friedman [ii] and Lee and Phillips {19] have concluded that there are significant differences between farm and
non.-
farm families, as well as other socioeconomic groups, in their
consumption response to changes in income.
Friedman estimated the marginal propensity to consume at
0. 50 and 0. 73 for farm and nonfarm families, respectively, using
1935-36 data. Based on 1941 data, the marginal propensity to con-
sume of farm families was about 0. 48 and that of urban families was
0. 79. According to Friedman, the observed differences in behavior
are consistent with his permanent income hypothesis.
The permanent income hypothesis can be stated by three
equations for the individual consuming unit:
2
= K (i,
(1. 1) C
W,
u) Y
(1.2)Y =Yp+Yt
(1.3)C =C+C
Equation (1. 1) asserts that the ratio (K) between permanent
consumption
(Ce) and permanent income
(Yr)
is independent of the
size of permanent income but does depend on other variables:
(1)
the rate of interest (i), (2) the ratio of nonhuman wealth to income
(w), and (3) other factors affecting the consuming unit's tastes and
preferences (u).
Equations (1.2) and (1. 3) are definitional. A consuming unit's
current income (y) and consumption (c) have permanent and transi.tory components. The permanent income of a consuming unit is
defined as the product of an interest rate (i) and the stock of wealth
(w); and the stock of wealth is interpreted as the present value of
anticipated future receipts from both human and nonhuman assets
discounted back to the present at a subjective rate of interest. The
permanent income is thus a theoretical construct. The amount by
which current income differs from permanent income is called
transitory income which reflects the influence of factors regarded
as chance or random by the consumer unit, as well as errors of
mea surement.
The permanent income hypothesis in the above form is incap-.
able of being tested empirically because neither permanent income
3
nor permanent consumption can be observed directly for an individual consuming unit. To make the hypothesis testable, Friedman
assumes that
(1.4)
=0
p
where p denotes the correlation coefficient between the variables
designated by the subscripts.
In addition to equation (1.4), it is also assumed that the mean
transitory component of both consumption and income are zero, or
(1.5)
0
With these assumptions, Friedman derived the following
equation for interpretation in terms of the permanent income hypothes is:
(1.6)
b
=K
(YY)2
(Y-Y)2
=KP y
where b is the marginal propensity to consume out of current income, P,. is the fraction of the total variance of income in the group
contributed by the permanent income, and K is as defined in equation (1. 1).
In his budget studies, Friedman, observing that farm
families have a lower marginal propensity to consume and a lower
value of
than do nonfarm families, concluded that the entre-
prenurial nature of farmer's income has greater uncertainty than
do other types of income. Furthermore, farm families were
expected to have a lower K because of their lower average income.
Thus, the differences in the consumption responses to changes in
income between farm and nonfarm families could be explained by
the permanent income hypothesis.
In 1 971, Lee and Phillips conducted a study to test the validity
of Friedman's permanent income hypothesis in explaining the differences in consumption behavior between farm and nonfarm fam-
ilies. Their findings indicated that the level and the stability of
income were not important contributing factors to observed differences in behavior. In their study, it was assumed that the perman-
ent income hypothesis is applicable for the different categories of
consumption.
The model used by Lee and Phillips was estimated by two
stage least squares. The model employed the assumption that the intercept and other regression coefficients are the same for farm and
nonfarm samples. This assumption may result in biased estimates
of the slopes, if in fact there are different intercepts for both groups.
In addition, group mean data were used in their analysis. Freund
[ii] indicated the the use of group means for performing regression
analysis is inferior to the use of all observations because of low
efficiency and precision..
In summary, the results of previous studies, while of interest,
are unsatisfactory because of conflicting findings in establishing the
5
permanent income hypothesis as an explanation of the differences
in consumption patterns between farm and nonfarm families.
Furthermore, neither of these studies has analyzed the factors
affecting K--the ratio of permanent consumption to permanent income. This is an essential aspect of the permanent income hypoth-
esis for understanding the consumption behavior of consumer units.
Thus, observed differences in the marginal propensities to consume
between farm and nonfarm groups still require further investigation.
Hence, the main objective of this study is to determine the
factors affecting the differences in the marginal propensities to
consume of farm and nonfarm families. To achieve this objective,
individual observation data from the Survey of Consumer Expenditure
were chosen for the study because it contains detailed information of
socioeconomic characteristics for the sample households.
The plan of this study is as follows: Chapter II deals with a
theoretical framework. Chapter III is devoted to a discussion of
the model and statistical procedure. Chapter IV is concerned with
the statistical results. Finally, Chapter V gives conclusions and
the implications of the study.
II THEORETICAL FRAMEWORK
This chapter deals with the theoretical basis of consumption
analysis. There are two sections. In the first section, factors
affecting the marginal propensity to consume are identified in the
context of the theory of consumer choice under constraint. The
second section is devoted to formulating hypotheses.
Consumer Choice and the Marginal
Propensity to Consume
The assumptions of consumer choice are (1) a consumer has a
given utility function which specifies his preferences for various
goods and services, (2) a consumer is subject to a budget restraint
in his choices of goods and services, and (3) the consumer attempts
to allocate his limited budget among available goods and services
so as to maximize his utility or satisfaction.
According to these assumptions, consumers maximize a
utility function of the form
U = Y (X1, X2,
(2.1)
.
.
Xni Xn)
subject to the budget constraint.
Y=
(2.2)
where X1, X2,
.
.
P.X.
11
X_q represent the consumption levels of the
n-i commodities, Xn represents the level of money balances,
7
P. are their prices, and Y is income.
However, this study deals with consumer choice between the
aggregate of all commodities and money balances. The aggregate
of all commodities can be denoted as q1
=
(X1, X2,
.
.
., Xi);
money balances are q2 = X. Thus, the consumer's choices are
limited to q1 and q2. The consumer's utility function can be ex-
pressed as:
(2. 3)
U=
q1, q2)
The consumer's budget constraint can be written as follows:
f(2.4)
Where P
Y = Pq1 + q2
(P1, P2.
.
.
., P1) represents a price index and the
price of q2, money balances, is always equal to one.
To maximize the utility function subject to the budget constraint the consumer should find a combination of commodities and
money balances that satisfies equation (2. 4) and also maximizes
equation (2. 3). One can construct the function:
(2. 5)
L=
(q1,
q2) + X (Y - Pq1 - q2)
where X is a Lagrangean multiplier. Maximizing L requires that
the first and the second order conditions be satisfied.
The first order conditions are that both partial derivatives
equal zero:
(2.6)
ci
(2.7)
= XP
X
2
0
=0
Thus, one obtains the first order condition from equations
(2. 6)and
(2.7)
as
ci
(2.8)
=p
This is a necessary condition for a maximum, but it does not
ensure that a maximum is actually reached. The second order condition must also be satisfied.
The second order condition requires:
-
(2. 9)
2P21 +P2
dq12
22
<0.
Taking the total derivative of the first order condition, one
would obtain an equation as follows:
dq1 +
-
(2. 10)
-
Pct
dq2 = 0.
Equation (2. 10) can also be written as:
(2. 11)
dq
2l
11
(
)
dq1
Taking the total derivative of the budget constraint, the
equation obtained is:
(2. 12)
dY = Pdq1 + dq2.
1"
Substituting equation (2. 12) into equation (2. 11), gives:
dq1
(2. 13)
22
12
dY
- 2P21
Thus,
dq1P
(2.14)
dY
PI)22
12
=
2zi +
i1
dq1P
where
dY
is the marginal propensity to consume which is de-
fined as the ratio of the change in consumption expenditure to the
change in income. Note that the right-hand side of equation (2. 14)
consists of the price index and elements from the utility function of
the consumer. Therefore, factors affecting the marginal propensity
to consume can be written in the functional relationship as
dq1P
(2. 15)
dY
=
lIP,
4)
°'i' q2)]
Thus, the marginal propensities to consume of two consumers
will differ if (1) their utility functions are different; or (2) the
prices which they pay are different; or (3) both.
Specifically, if two consumers have linear consumption func-
tion such as:
(2. 16)
C.I = a. + 3. Y.
I
1
1
Where C. is the expenditure for current consumption of the ith
consumer, Y. is his income, a. is an intercept, and 1 is his
10
marginal propensity to consume,
13.
will differ between consumers
only if one of the above conditions is true.
The Formulation of Hypotheses
The conclusion reached in the preceding discussion is that the
marginal propensity to consume is determined by the consuming
unit's utility function and the price level. The prices paid by con-
suming units are usually not available in cross sectional data; the
failure to include price is, essentially, equivalent to assurrLing that
the price level is the same for all consuming units in the sample.
Furthermore, it is difficult to measure the consuming unit's utility
function empirically.
However, socioeconomic characteristics of consuming units
can be observed directly. Changes in socioeconomic characteris-
tics of consuming units generally contribute to changes in their
tastes and preferences. It is associated with the utility function.
Thus, the marginal propensity to consume will be a function of
socioeconomic conditions of the consuming units. The marginal
propensity to consume of two consuming units will differ if their
characteristics are different. This discussion suggests one testable
hypothesis concerning the marginal propensity to consume of farm
and nonfarm consuming units:
11
Hypothesis #1. Farm and nonfarm consuming units with honogeneous socioeconomic characteristics have the same
propensity to consume.
If this hypothesis is accepted, then observed differences in
the marginal propensities to consume for farm and nonfarm consuming units by the previous studies could be due to differences
in the distribution of consuming units by socioeconomic character-
istics in the two samples. This argument suggests one more testable hypothesis:
Hypothesis #2: Distribution of consuming units by socioeconomic
characteristics in the farm sample is different
from that of consuming units in the nonfarm sample.
The acceptance of these hypotheses will provide an explanation
for observed differences in the consumption behavior between farm
and nonfarm consuming units. It also provides evidence for the
prediction from the theory of consumer choice that the marginal
propensity to consume is determined by the utility function and the
price level.
12
III THE MODEL AND STATISTICAL PROCEDURE
In the preceding chapter a theoretical framework was presented. The present chapter is an attempt to derive the model and
the statistical procedure for estimating the model and testing the
hypotheses. These are presented below.
mL.
In Chapter II, it was argued that the marginal propensity to
consume is a function of utility function which is, in turn, assumed
to be a function of the socioeconomic characteristics of the consuming unit. This can be expressed as below:
(3. 1)
p
= f (Xi,
.
.
.
,
X
n
where p is the marginal propensity to consume and X1, X2,
.
Xncorrespond to the socioeconomic characteristics of the consuming unit. The functional relationship of (3. 1) may be explicitly
written as;
(3. 2)
where p0, p1,
p0 + p1X1 + p2X2 +
p
.
.
.,
p
+
X
are parameters. Substituting equation
(3. 2) into a consumption function, i. e., C = a + pY, it becomes
(3.3)
XY
X Y+p22
X Y+. . . +p nfl
C = a+p 0Y+ 11
where C is expenditure for current consumption and Y is disposable
13
income.
Equation (3. 3) takes factors affecting the marginal
propensity to consume into consideration. However, the estimate
of the intercept (a) is also affected by characteristics of the consuming units because a is a function of p
i. e., a = C - pY.
Thus, the consumption function can be rewritten as:
(3.4)
C = A0 +A11
X +.
+ 13 Y +
0
1
X Y.
11
.
.
. +A nn
X
.
nn
+ p x Y
Equation (3.4) is an econometric model used in this study.
The Statistical Procedure
The first problem the researcher encounters in estimating
equation (3. 4) is the selection of the independent variables (X1, X2,
Xn) to be included in the equation. The variables selected in
this study include (1) size of the consuming unit, (2) education
level, (3) occupation, (4) age, (5) regionality, and (6) urbanization.
The selection is based on a priori reasoning and available data,
since it is neither practical nor possible to include in the model all
of the socioeconomic characteristics which may influence the taste
and preferences of the consuming unit.
The reasons for not using permanent income are (1) t1is
study is not testing the permanent income hypothesis, and (2) it is
very difficult to quantify and measure the term empirically.
14
Since the variables selected are classification variables
rather than continuous variables, it is difficult to quantify them.
To overcome this obstacle, dummy variables are used to measure
them. Each variable takes a value of one or zero, depending upon
whether or not the observation is in the particular class represented
by that dummy variable. In order to avoid the problem of over-
identification which is created by perfect correlation involved in a
set of dummy variables which are mutually exclusive, a constraint
must be placed on the system. This constraint is to drop one of the
dummy variables from each mutually exclusive system. The re-
maining variables would no longer be perfectly correlated and the
solution for the parameters would be obtainable.
The ordinary least squares method is then employed to estimate the regression coefficients of the consumption function. But
the selected variables have about 32 classes with more than ten
thousand observations which is over the capacity of available com-
puting facilities. So the first step is to attempt to reduce the number
of variables involved. The method adopted is to run regression
analyses for each socioeconomic variable separately.
The regression equation for each socioeconomic variable is
presented below:
15
Urbanization equation:
(3.5)
2
C = A uo
+
i=1
2
Aui.U.+ p uo Y+
1
i=1
piii.U.Y+E U
Regionality Equation:
3
(3.6)
C = A ro
+
i=1
3
Ari.R.+proY+
1
i=1
pri.R.Y+E r
1
Education equation:
(3. 7')
3
C = A eo
3
Aei.E. + p eo Y + E p ei.E.Y + E
+
i-i
1
1
1=1
e
Age equation:
5
(3.8)
C = A ao
5
+
IZ1
Aei.A. + p ao Y +
1
i=1
pai.A.Y + E a
1
Size equation:
6
(3.9)
C = A SO
6
+
i=1
ASi.S. + p so Y + E pSI.S.Y + E S
1
1
i=1
Occupation equation:
6
(3.10)
6
C=A00+A.O.+p 00Y+p.O.Y+E
01
1
i=l
where
01
1
0
C = expenditure for current consumption
Y = disposable income
u1 = 1, if the consuming unit lies in rural nonfarm
sample; = 0, otherwise.
16
u2
R1
=
1,
if the consuming unit lies in urban sample;
=
0,
otherwise.
=
1,
if the consuming unit lies in the Northcentral
region;
R2
= 1,
=
0,
=
0, otherwise.
if the consuming unit lies in Southern region;
otherwise.
R3 = 1, if the consuming unit lies in Western region;
= 0, otherwise.
1,
if 8 years or less of education;
=
0, otherwise.
0, otherwise.
E2
=
1,
if 9-12 years of education;
E3
=
1,
if 13-16 years of education;
A1
=
1,
if <24 years old;
A2
=
1,
if 25-34 years old;
A3
= 1,
if 35-44 years old;
A4
=
1,
if 45-54 years old;
=
0, otherwise.
A5
= 1,
if 55-64 years old;
=
0, otherwise.
S1
=
=
=
0,
=
=
0, otherwise.
otherwise.
0, otherwise.
0, otherwise.
1,
of 2.0-2.9 persons in the consuming unit;
0,
otherwise.
S2 = 1, if 3.0-3.9 persons in the consuming unit;
= 0, otherwise.
S3 = 1, if 4. 0-4. 9 persons in the consuming unit;
= 0, otherwise.
S4 = 1, if 5. 0-5.9 persons in the consuming unit;
17
= 0, otherwise.
S5 = 1, if 6. 0-6. 9 persons in the consuming unit;
0, otherwise.
= 1, if 7. 0-7. 9 persons in the consuming unit;
= 0, otherwise.
01 = 1, if self-employed;
=1
0, otherwise.
if professional and managers; = 0, otherwise.
03 = 1, if clerical and sales;
0,
otherwise.
04 = 1, if skilled workers; = 0, otherwise.
05 = 1, if semi-skilled workers; = 0, otherwise.
= 1,
E
if unskilled workers; = 0, otherwise.
= the error term.
Once the regression coefficients for the above equations are
obtained, one can then derive the marginal propensity to consume
(MPG) for each class of the selected variables. For instance, the
estimated urbanization equation is as follows:
A
(3.11)
G =A uo
A
A
+A.0
+Au2U
ui
2 +puoY+.UY+
ui
1
1
A
uZUY
2
Then, the equations for each urbanization class can be derived from
equation (3. 11) as follows:
Farm consuming unit:
(3.12)
A
G =Auo
A
+13
uo
Urban consuming unit:
EI]
(3.13)
A
C =(A uo +AUi)+(1A UO +p Ui.)Y
A
Rural nonfarm consuming unit
(3.14)
C =(A uo +Au2)+(p uo
A
+pu2)Y
A
Thus, the marginal propensity to consume obtained from
equations (3. 12), (3. 13), and (3. 14) for farm, urban, and rural
A
nonfarm consuming units are 3uo ,
p
uo + p ui., and 13 uo
A
+ P
uZ
respectively. An interesting feature about the regression coefficients in equations (3. 12), (3. 13) and (3. 14) is that they are ex-
actly the same as those that would be obtained from three separate
regressions of C on Y, one estimated from farm observations, and
the other from urban and rural nonfarm observations.
A signifi-
cant advantage of using the dummy variable regression technique
over the separate independent regression technique is that the
number of degrees of freedom is considerably greater permitting
more powerful tests of significance.
To determine if the marginal propensity to consume of one
urbanization group (MPC.) is different from that of the other group
(MPG.), one must determine whether or not there is a significant
difference between MPG. and MPG.. The t-test is used;
1
J
For proof, see AppendixA.
19
MPG. - MPG.
1
(3.15)
S2
i
-MPG)
If the value of this statistic is significantly different from
zero, the hypothesis that MPG. = MPG. is to be rejected. The
results of the t-tests for each class of the variables selected provide a basis upon which to group several classes together, thereby
reducing the number of variables involved in the model.
The next step is to introduce those explanatory variables
identified from the results obtained in the above process into the
dummy variable regression model in equation (3. 4) for farm and
nonfarm consuming units. The MPG's for farm and nonfarm con-
suming units with respect to a particular set of homogenous char-
acteristics can be obtained from the estimated regression equation.
These MPG's are then subjected to the t-test. This is a test of
the hypothesis that there is no significant difference in the marginal
propensity to consume for farm and nonfarm consuming units when
they have homogeneous socioeconomic characte ristics.
To test the hypothesis that there is a significant difference in
the distribution of consuming units by socioeconomic characteristics
for farm and nonfarm samples, the consuming units were first
grouped according to socioeconomic characteristics. This tabular
analysis provides a measure of the distribution of consuming units
20
in the two samples. The percentage of consuming units with
certain types of socioeconomic characteristics studied is then
calculated for each sample. Finally, the t-test is used to deter-
mine statistical significance of difference in this percentage between
farm and nonfarm samples.
21
IV EMPIRICAL RESULTS
This chapter begins by discussing the source of the data and
the terms used in this study, thus providing the background for understanding the subsequent discussion of analyses. It is then devoted
to discussing statistical results: (1) the marginal propensity to consume out of disposable money income, and (2) the marginal propensity to consume out of total disposable income.
ffil-..
T\-4-.
The data employed in the following analysis was obtained from
the Bureau of Labor Statistics and the U.S. Department of Agriculture--Survey of Consumers Expenditure Conducted in 1960-61. A
three-stage sample design was used to select a sample representing
families and single consumers in the population.
The data were
collected by personal interviews. The survey contained 13, 728
observations.
For this study, only selected observations within the total
sample were considered. The selection was based on the following
criteria:
3' For detail on sampling method, see [26, p. 11-16].
22
(1) Income is in the range of $0 to $15, 000.
Those observations not in this range were excluded on the grounds
that they represented extreme cases.
(2) Only households of two or more persons were considered.
Those households with average annual family size of less than two
persons are comprised of bachelors, unmarried women, the
divorced, widows and widowers. This is a very heterogeneous
group, and thus they were eliminated from the analysis.
(3) Those observations where family size, education of head,
age of head, and occupation of head were not reported were excluded.
This was necessary because these observations could not be analyzed due to lack of information concerning relevant variables.
Based on the above criteria, 10, 218 families were selected
for this study. In the empirical analysis of this study, the family is
considered to be the consuming unit. However, as discussed in the
previous chapters, a consuming unit may be either an individual
consumer or a family.
The Definitions
Family
The family is defined as two or more persons dependent on a
common, or pooled, income for their major items of expenditure,
23
and usually living in the same household.
Family Size
It is the number of equivalent full-year members, based on
the total number of weeks during which both full and part-year mem-
bers belonged to the family in the survey year. Fifty-two weeks of
family membership are considered equivalent to one person, et
cetera.
Family Head
In husband-wife families, the husband is always considered
the head. In other types of families, the person recognized as the
head by other family members is so designated.
Education of Family Head
The number of years of study completed by the family head by
the end of the survey year in schools which advance a person to an
elementary or high school diploma, or to college, university or professional school degree.
Age of Family Head
It is that recorded as the age in years of the family head at
the end of the survey year.
24
Occupation of Family Head
It is based on the family head's major occupation, i.e., the
occupation at which the family head was employed for the greatest
number of weeks in the survey year.
Urban Families
Families that reside in incorporated places of 2, 500 population
or more and in the densely settled areas immediately adjacent to
cities of 50, 000 population or more.
Rural Nonfarm Families
Families that reside outside of urban areas, but not on farms.
Rural Farm Families
Families that reside outside of urban areas, and on a farm,
defined as in the 1960 Census as a place of 10 acres or more from
which the sale of crops, livestock products, et cetera, amounted to
$50 or more; or a place of less than 10 acres with sales of $250 or
more.
25
Northeast Region
It includes the states of Maine, New Hampshire, Vermont,
Connecticut, New York, Pennsylvania, and New Jersey.
Nortbcentral Region
This region consists of Ohio, Indiana, Illinois, Michigan,
Wisconsin, Minnesota, Iowa, Missouri, North Dakota, South
Dakota, Nebraska, and Kansas.
Southern Region
It includes Maryland, Virginia, West Virginia, North Caro-
lina, South Carolina, Georgia, Florida, Kentucky, Tennessee,
Alabama, Mississippi, Arkansas, Louisiana, Oklahoma and Texas.
Western Region
This includes Montana, Idaho, Colorado, New Mexico, Utah,
Washington, Oregon, California, Wyoming, Nevada, and Arizona.
Money Expenditure for Current Consumption
These expenditures consist of cash outlays for goods and
services for family living during the survey year.
Total Expenditures for Current Consumption
It includes cash and noncash outlay for goods and service for
family living.
Disposable Money Income
It has been measured by total money income after deduction
of personal taxes.
Total Disposable Income
It is total money and nonmoney income after taxes.
The Marginal Propensity to Consume Out of
Disposable Money Income
The marginal propensity to consume out of disposable money
income is defined as the ratio of the change in money expenditures
for current consumption to the change in disposable money income.
This can be obtained from the slope of the linear regression of
money expenditure (C1) on disposable money income (Y1). The
statistical procedure used in this regression analysis has been discussed in Chapter III. The results of the empirical estimation of
the coefficients for each socioeconomic variable are presented
below.
27
The regression coefficients of the urbanization equation in
Table 1 are computed by least square multiple
regression.'
The
figures in the parentheses below the coefficients are the t-values.
All coefficients in equation (4. 1) are significantly different from
zero at the one percent level in a t-test. The estimate of the coefficient of multiple determination (R
2.is 0. 65.
)
This indicates that
about 65 percent of the variation in money expenditure is explained
by the variables used in. the analysis.
The equations for rural nonfarm families, urban families, and
rural farm families are derived from equation (4. 1). All coefficients
in equations (4. 2), (4. 3) and (4. 4) are significant at the one percent
level. -" Equation (4. 2) indicates that the marginal propensity to
1etroscedasLicity in the. regression model was not examined in the study because it would be too costly to calculate
residuals for 10, 218 observations in each regression equation.
No high degree of multicollinearity in the regression equations is
present by observing the correlation coefficient matrices. In
general, size of sample tends to infinity the problem of high degree
of multicollinearity decreases. The size of sample in this study
being 10, 218 is so large. Thus, the problem of high degree of
multicollinearity is not likely to occur.
The following was used to calculate a t-value for derived
equations:
a. + a.
I:
s.e. (a. +a.
1
3
1/2
where s. e. (a. + a)
[Var a. + Var a. + 2 Coy a.a.]
13
3
For a further discussion, see Kmenta [16] and Gujarati [13].
1
.1
1
Table 1. Regression coefficients and related statistics of urbnization equation in the U. S., 1961.
Constant term
Equation
U
1
U
2
Y
1
UY
1 1
UY
2 1
Urbanization:
1655.90
(22.32)**
(4.1) C1
-644.73
(-5.90)**
-435.04
(_4.94)**
0.4649
(33.21)**
Rural nonf arm families:
(4.2) C1
1011.17
(5.99)**
0.7255
(52.96)**
1220.86
0.7393
(108. 72)**
Urban families
(4.3) C
(7.84)**
Rural farm families:
1655.90
(4.4) C1
(22. 32)**
where
money expenditure for current consumption
C1
= disposable money income
U1
U2 =
1,
if the rural nonfarm families; = 0, otherwise
1,
if the urban families; = 0, otherwise
** Significant at the 1% level.
0. 4649
(33. 21)**
0.2606
(13.33)**
0.2744
(17.64)**
0.65
29
to consume out of disposable money income is 0. 4649 for farm
families. Thus, an increase of disposal money income by one
dollar would bring an increase in money expenditures of about 46
cents in the farm families. Similarly, equations (4. 3) and (4. 4)
indicate that the marginal propensities to consume are 0. 7255 and
0. 7393 for rural nonfarm and urban families, respectively.
The results of testing the differences in the marginal propen-
sities to consume for farm, rural nonfarm, and urban families are
given in Table 2. The test provides evidence that the marginal pro-
pensity to consume for farm families is significantly different from
that of rural nonfarm families and urban families. However, there
is no significant difference in the marginal propensity to consume
between rural nonfarm and urban families. Thus, these findings
support Friedman's conclusion that farm and nonfarm families have
different marginal propensities to consume.
The regression coefficients presented in equations (4. 6), (4. 7),
(4. 8), and (4. 9) in Table 3 are significant at the one percent level.
The marginal propensities to consume are about 0. 7477 and 0. 7305
for the Northeast region and the Southern region, respectively,
while that of the Northcentral region and the Western region are
about 0. 6714 and 0. 6824, respectively. The results of testing the
differences in the marginal propensities to consume show that the
marginal propensity to consume of the Northeast region is not
30
Table 2. Differences in the marginal propensities to consume and
their t-values for farm, rural, nonfarm, and urban
families in the U.S., 1961.
Rural nonfarm
Rural nonfarm
Urban
Farm
0
(0)
Urban
Farm
-0. 0138
0
(-0.91)
(0)
0. 2606
(13. 33)**
0. 2744
(17. 64)**
0
(0)
** Significant at the 1% level.
significantly different from that of the Southern region. Similarly,
there are no significant differences in the marginal propensities to
consume between the Northcentral region and the Western region
(Table 4).
The above evidence reveals the existence of regional variations
in the marginal propensity to consume. These variations may re-
flect regional differences in (1) cost of living, (2) customs, and (3)
tastes conditioned to climate.
Educational attainment of the family head presented as a
variable in the regression analysis appears to be related to the
family's consumption patterns (Table 5). The marginal propensity
to consume is lower for families whose head had been in college or
in graduate school. Table 6 shows that the marginal propensity to
Table 3. Regression coefficients and related statistics of regionality equation in the U. S., 1961.
Equation
Constant term
R1
R2
R3
1247.72
(14.90)**
18.59
-299.06
(_2.93)**
435. 12
(0.17)
Y1
R1Y1
R2Y1
R3Y1
Regionality equation:
(4.5) C1
0.7477
(3.30)** (62.08)**
-0. 0763
-0. 0172
-0. 0653
(-4.81)**
(-1.09)
(_3.55)**
0.63
Northc entral region:
(4.6) C1
1266.30
(18.66)**
0.6714
(25.63)**
948.66
(5.3S)**
0.7305
(27.99)**
1682.84
(16.49)**
(24. 55)**
1247.72
(14.90)**
0.7477
(62.08)**
Southern region:
(4.7) C1
Western region:
(4.8) C1
0.6824
Northeast region:
(4.9) C1
where
=
Y1
money expenditure for current consumption
disposable money income
=
R2
1,
if Northcentral region;
1,
if Southern region;
0, otherwise
0, otherwise
R3 = 1, if Western region; = 0, otherwise
**
Significant
at the 5% level.
I-'
32
Table 4. Differences in the marginal propensities to consume and their t-values for four regions
intheU.S., 1961.
Northc entral
Northcentral
Southern
Western
Northeast
0
(0)
Southern
Western
-0. OSO 1
0
(4.28)**
(0)
-0.0110
(0.63)
Northeast
0.0481
(2.86)**
-0. 0763
-0. 0172
(-4.81)**
(-1.09)
0
(0)
-0. 0653
(_3.55)**
0
(0)
** Significant at the
consume for eight years or less of education of the family head is
not different from that of 9 or 12 years education of the family head.
The effect of the educational attainment of the family head on
the marginal propensity to consume may reflect the tastes of the
consumption unit. Burk's study [4, p. 121-122] indicated that
household heads with a higher educational attainment had a larger
amount of expenditures on education and reading. The investment
in education for the family's members is usually expensive and over
a limited number of years. Thus, it requires more saving for that
purpose. In addition, it is in general far easier to borrow on the
basis of a tangible physical asset than on the basis of human capital
such as education. Therefore, the families whose head have a
higher educational attainment may need an additional reserve. This
Table 5. Regression coefficients and related statistics of education
equation in the U. S., 1961.
quation
Constant term
Education equation
(4. 10) C1
2435.35
(9.76)**
E1
E2
-1481.23 -1067.28
(-5.80)** (-4.16)**
E3
-546.90
(-2.00)*
0.5924
(21.33)**
E1Y1
E2Y1
E3Y1
0.1130
(3.81)**
0.1086
(3.72)**
0.0683
(2.19)*
R2
0.63
8 years or less:
(4.11) C
1
954.12
(1.90)
0_70S4
(67. 18)**
1368.07
(2.72)**
0.7010
(77.89)**
1888.45
(3.69)**
0.6607
(46.53)**
2435.35
(9.76)**
0.5924
(21.33)**
9 to 12 years:
(4.12) C1
13 to 16 years:
(4.13) C1
17 years or more:
(4.14) C1
where
= money expenditure for current consumption
= disposable money income
E1 =
1,
if 8 years or less of education; = 0, otherwise
E2
1,
if 9 to 12 years of education; = 0, otherwise
E3
1,
if 13 to 16 years of education; = 0, otherwise
** Significant at the 5% level
*
Significant at the 1% level
tj
U)
34
Table t5. Differences in marginal propensities to consume and their t-values for four educational
levels of family head in the U.s., 1961.
8 years
or less
8 years or less
9-12
years
13-16
years
17 years
or more
0
(0)
9-12 years
13-16 years
17 years or more
0.0044
(0.32)
(0)
0.0447
(a.S4)*
0.0403
(2.40)*
0
0. 1130
0. 1086
(3.81)**
(3.72)**
0
(0)
0.0683
(2.19)*
0
(0)
** Significant at the
* Significant at the
may contribute to their low marginal propensity to consume.
From equation (4. 16) to equation (4. 21), one may get the im-
pression that the marginal propensity to consume decreased in gen.-
eral as the age of the family head increased (Table 7). The pattern
observed probably reflects the fact that older families are more
"stocked up" with durables. It seems that there is a tendency for
younger families to make heavy purchases of durable goods even
though they may have to dissave to do so. On the other hand, older
families with necessary assets, may make relatively few durable
goods purchases.
Table 8 indicates that there are no significant differences in
the marginal propensities to consume among the families whose
head is in the age ranges: less than or equal 24 years, 25 to 34
Table 7. Regression coefficients and related statistics of ages equation in the U. S., 1961.
Constant
term
Equation
A1
A2
A3
A4
A5
Y1
A1Y1
A2Y1
A3Y1
A4Y1
A5Y1
Age equation:
(4.15) C1
1038.43 143.42 366.35
(12.03)** (0.64) (2.89)**
343.39
86.84
(2.92)** (0.74)
88.93 0.6102
0.1536 0.1035 0.1082 0.1191 0.0547 0.64
(0.74) (33.35)** (3.34)** (4.39)** (5.07)** (5.56)** (2.43)*
Less than 24 years old:
(4.16) C1
1181.82
(5. 73)**
0.7638
(18. 06)**
25 to 34 years old:
(4.17) C1
1404.78
(15. 08)**
0.7137
1381.82
(17.34)**
0.7184
(65.31)**
1125.27
(14.04)**
0. 7293
(65.12)**
1127.36
(13.50)**
0. 6649
(51. 15)**
1038.43
0.6102
(33.35)**
(47. 90)**
35 to 44 years old:
(4.17) C1
45 to 54 years old:
(4. 10) C1
55 to 64 years old:
(4.20) C1
More than 65 years old:
(4.21) C1
(12.03)**
where
C1 = money expenditure for current consumption
Y1 = disposable money income
= 1, if less than 24 years old; =
0, otherwise
A1
= 1, if 25 to 34 years old; =
otherwise
0,
A2
= 1, if 35 to 44 years old; =
0, otherwise
A3
A
A
=
**
Significant at the 1% level
Significant at the 5% level
*
=
1 if 45 to 54 years old; = 0, otherwise
1 if 55 to 64 years old; = 0, otherwise
U.)
U-'
Table 8. Differences in the marginal propensities to consume and their t-values for six age groups in the U. S., 1961.
24 years
or less
24 years or less
25-34
years
35-44
years
45-54
years
55-64
years
Over
65 years
0
(0)
25-34 years
0.0501
(1.12)
35-44 years
45-54 years
55-64 years
Over 65 years
0
(0)
0. 0454
0. 0047
(1.04
(0.25)
0
(0)
0.0345
0.0156
0.0109
0
(0.79)
(0.84)
(0.69)
(0)
0.0989
(2.24)*
0. 0488
(2.47)*
0.0535
(3.13)**
(3.74)**
0
(0)
0. 1535
0. 1035
0. 1082
0. 1191
0. 0547
0
(3,34)**
(4.39)**
(5.07)**
(5.56)**
(2.43)*
(0)
0. 0644
** Significant at
*
Significant at
C'
37
years, 35 to 44 years, and 45 to 54 years in age. However, the
marginal propensities to consume for the families with age of the
family head in the ranges of 55 to 64 years, and 65 years old or
more, are significantly different from each other.
The marginal propensity to consume is the lowest for families
with 2. 0 to 2. 9 persons and is the highest for the largest families
with 8. 0 or more persons. This gives an indication that the family
with a larger number of members would tend to have a greater
marginal propensity to consume (Table 9).
Crockett and Friends [9, p. 72-92] described the effect of
family size as follows: (1) Most of the effect of family size on
consumption expenditures is reflected in changes in expenditure on
necessary items such as food, clothing, et cetera, and (2) total
consumption goes up markedly as family size increases. Thus, the
high marginal propensity to consume for large families is a result
of total consumption increases as family size increases. With an
increase in expenditures on relatively necessary goods, the family
cannot but decrease expenditures on other commodities and services.
However, this is not easy to achieve because the consumption habit
is relatively rigid. Therefore, a solution to this is to reduce their
saving.
It is then observed from Table 11 that the lowest marginal
propensity to consume occurred in families with a self-employed
Table 9. Regression coefficients and related statistics of family size equation in the U. S., 1961.
Equation
Constant
term
-
F1
F2
F3
F4
F5
F6
F1Y1
F2Y1
F3Y1
F4Y1
F5Y1
F6Y1
R
2
Family size
equation:
(4.22) C1
1010.58 101.66 278.06 436.99 420.96 406.01 582.25 0.7530 -0. 1093 -0.0451 -0.0393 -0.0280 -0.0186 -0.0621 0.64
(S.31)** (0.51) (1.35) (2.08)* (1.89) (1.68) (2.07)* (25.67)**(_3.52)** (-1.42) (-1.23) (-0.84) (-0.52) (-1.50)
2.0 to 2.9
persons:
(4.23) C1
1112.25
(19.50)**
0.6437
(10.82)**
1288.64
(16.31)**
0.7079
(11.82)**
1447.57
(16.05)**
0.7137
(11.90)**
1431.54
(12.29)**
0.7250
(11.91)**
1416.S9
0.7344
(11.79)**
3.0 to 3.9
persons:
(4.24) C1
4.0 to 4.9
persons:
(4.25) C 1
5.0 to 5.9
persons:
(4.26) C1
6.0 to 6.9
persons:
(4.27) C1
(9.44)**
Continued
Table 9--Continued.
Constant
term
Equation
F1
F2
F3
F4
F5
F6
Y1
F1Y
F2Y1
F3Y1
F4Y1
F5Y1
F6Y1
R2
7.0 to 7.9
persons:
1592.83
(7. 70)**
(4.28) C1
0, 6909
(10. 53)**
More than
8.0 persons:
1010.58
(5. 31)**
(4.29) C1
where
money expenditure for current consumption
1
F1
F2
F3
F4
F5
0. 7530
(25. 67)**
=
disposable money income
=
1,
=
1,
=
1,
=
1,
if 40 to 4.9 persons in the family; = 0, otherwise
if 5.0 to 5.9 persons in the family; = 0, otherwise
=
1,
if 6.0 to 6.9 persons in the family;
if 2.0 to 2.9 persons in the family; = 0, otherwise
if 3.0 to 3.9 persons in the family; = 0, otherwise
= 0,
otherwise
= 1,
if 7.0 to 7.9 persons in the family; = 0, otherwise
** Significant at the 1% level
*
Significant at the 5% level
F6
0
Table 10. Differences in the marginal propensities to consume for seven family size groups in the U. S., 1961.
2.0-2.9 persons
2.0-2.9
3.0-3.9
4.0-4.9
5.0-5.9
6.0-6.9
7.0-7.9
Over 8.0
persons
persons:
persons
persons
persons
persons
persons
0
(0)
3.0-3.9 persons
4.0-4.9 persons
5.0-5.9 persons
6.0-6.9 persons
-0.0642
(_4.12)**
-0. 0700
(_4.32)**
(-0.33)
0
(0)
-0. 0813
-0. 0172
-0. 0113
(-0.85)
(-0.54)
0
(0)
-0. 0907
-0. 0436
-0. 0207
0. 0094
0
(-1.80)
(-0.84)
(0.35)
(0)
0. 0170
0. 0228
0. 0341
0. 0435
0
(0.54)
(0.71)
(1.01)
(1.20)
(0)
-0.0451
-0. 0393
-0. 0280
-0. 0186
-0. 0621
0
(-1.42)
(-1.23)
(-0.84)
(-0.52)
(-1.50)
(0)
-0. 0472
(-2.42)*
Over 8 persons
-0. 0058
(_4.25)**
(-3.88)**
7. 0-7.9 persons
0
(0)
-0. 1093
(-3.52)**
** Significant at 1% level
* Significant at 5% level
0
Table 11. Regression coefficients and related statistics of occupation
equation in the U.S., 1961.
Equation
Constant
term
01
02
03
04
O
°6
1
01Y1
02Y1
03Y1 04Y1
05Y1
06Y1
R2
Occupation equation:
(4.30) C1
894.18 734.11 770.07 9.63 213.56
(9.00)** (6.03)** (5.00)**(0.06) (1.43)
162.78 -5.68 0.6950
-0.1440 -0. -120 0.1136 0.0519 0.0590 0.0884 0.64
(1.12) (-0.42)(30.73)** (-5.66)** (-0.45) (3.74)** (1.86) (2.06)* (3.05)**
Self-employed:
(4.31) C1
1628.29
(23.18)**
0.5510
(11.80)**
1664.25
(14.11)**
0.6830
(14.44)**
903.81
(6.41)**
0.8086
(39.83)**
Professional and mgrs.
(4.32) C 1
Clerical and sales:
(4.33) C1
Skilled workers:
(4.34) C1
1107.73
(9.91)**
0.7469
(45.82)**
Semi-skilled workers:
(4.35) C1
1056.95
(8.89)**
0.7540
(43.09)**
888.50
(4.05)**
0.7834
(43.28)**
Unskilled workers:
(4.36) C1
Continued
Table 11--Continued.
Equation
Constant
term
01
°2
03
04
0
°6
y1
01y1
Retired people:
(4.37) C1
894.18
0. 6950
(30. 75)**
(9. 00)**
where
money expenditure for current consumption
disposable money income
0
02 =
03 =
1,
if self-employed; = 0, otherwise
1,
if professional and managers; = 0, otherwise
1,
if clerical and sales; = 0, otherwise
04
1,
if skilled workers; = 0, otherwise
0
1, if semi-skilled workers; = 0, otherwise
1,
if unskilled workers;
** Significant at the 1% level
* Significant at the 5% level.
0, otherwise
02Y1
03Y1
04Y1
05Y1
06Y1
R2
43
family head. The self-employed families save a substantially
high proportion of their income as compared to other occupational
classes. This is true for the self -empl oyed in general. Ownership of a business or professional practice brings about a strong
need for investment funds thus resulting in a high level of saving.
Furthermore, the income of the self-employed family is more
uncertain than that of other occupational classes. Thus, more
money has to be saved for emergency use according to Friedman.
The marginal propensity to consume of self-employed families was significantly different from that of other occupational groups
(Table 12). This is similar to results reported by Friedman [ii]
and by Klein [16]. Friedman estimated that the marginal propensity
to consume for independent business was 0. 54 but it was about 0. 82
for others, using 1948 -50 data. Klein reported, in the 1950 and
1951 survey data for the U.S., the marginal propensity to save for
farmers and business men together was about 0.40-0.45. For nonfarmers and nonbusiness men units, the marginal propensity to save
was 0. 20.
The foregoing analyses indicates that some marginal propensities to consume with respect to socioeconomic variables of fam-
ilies are not significantly different from each other. This provides
a basis on which to group them together. The next step is to
explicitly include them in a multiple regression model. The
Table 12. Differences in the marginal propensities to consume for seven occupational groups in the U. S., 1961.
Selfemployed
Self-employed
Professional
and managers
Clerical
Skilled
and sales
Semiskilled
Unskilled
Retired
0
(0)
Professional and managers
-0. 1320
(-7.29)**
Clerical and sales
Skilled workers
Semi-skilled workers
Unskilled workers
Retired
** Significant at 1% level
* Significant at 5% level
0
(0)
-0. 2576
-0. 1256
0
(-5.06)**
(-2.44)*
(0)
-0. 1959
-0. 0639
0. 0617
0
(3.96)**
(-1.28)
(2.37)*
(0)
-0. 2030
-0. -710
0. 0546
0. 0071
0
(_4.07)**
(-1.41)
(2.04)*
(0.30)
(0)
-0. 2324
-0. 1004
0.0252
-0. 0365
-0. 0294
0
(-4.64)**
(_1.98)*
(0.93)
(-1.50)
(-1.17)
(0)
0.0519
(1,86)
0.0590
(2.06)*
0. 0884
-0. 1440
-0. 0120
0. 1136
(-5.66)**
(-0.45)
(3.74)**
(3.05)**
0
(0)
45
multiple regression equation fitted to the data is presented below.
Farm families equation:
C =ao+a 11
R +a 21
E +a E
32+a A
41+aA52+a5S1
(4.38)
1
+aO
71 +aO
82 +aO
93 +bY
ol +bRY
111 +bEY
211
+ b3E2Y1 + b4AJY1 + b5A2Y1 + b6S1Y1 + b7O1Y1
+ b8O2Y1 + b9O3Y1 + e
Nonfarm families equation:
C1 = a
(4. 39)
+ a'1R1 + a'2E1 + a'3E2 + a'4A1 + a5A2 + a'6S1
+aTO
71 +a'O
82 +a10
01 +b'RY
93 +b'Y
111 +b'EY
211
+ b3E2Y1 + b'4A1Y1 + bt5A2Y1 + b'6S1Y1 + b'701Y1
+ bt802Y1 + b9O3Y1 + e
where
C1
= Money expenditure for current consumption.
Y1
= Money disposable income.
R1
=
1,
if Northcentral and Western regions; = 0, otherwise.
=
1,
if 13 to 16 years of education; = 0, otherwise.
1
if more than 17 years of education, = 0, otherwise.
E2 =
A1 =
1, if the age of family head is 55 to 64 years; = 0,
otherwise.
A2 =
if the age of family head is more than 65 years; = 0,
otherwise.
S1
1,
=
1,
if 2. 0 to 2. 9 persons in the family; = 0, otherwise.
46
0
1,
if self-employed; = 0, otherwise.
02 =
1,
if professional, manager, or retired; = 0, otherwise.
1,
if clerical, sales, or unskilled worker;
03 =
0,
otherwise.
e
= a random error term.
The results obtained using the above equations are given in
Table 13. The impact of the socioeconomic variables of the family
on the marginal propensity to consume is discussed below.
Farm families;
The negative coefficient of the regional variable (-0. 2033)
shows that families in the Northcentral and Western regions have
about 0. 2033 lower marginal propensity to consume than farm fam-
ilies residing in the Northeast and Southern regions. If the family
head has a college education, the marginal propensity to consume
would be decreased about $0. 1340. The negative value of the family
size variable indicates that as the family size decreased to 2. 0 to
2. 9 persons, there is a tendency for the marginal propensity to
consume to decrease. The coefficient for the self-employed vanable is -0. 2076. This indicates that the shift from skilled or semi-
skilled worker to a self-employed occupation would tend to decrease
the marginal propensity to consume by $0. 2076.
47
Table 13.
Multiple regression coefficients and related statistics in farm and n
U.S., 1961.
Regressors
Farm
Nonf arm
families
fanulies
Constant
885.93
(3.02)**
1137.60
(13.08)**
R
1022.75
(6.90)**
(1.28)
E1
1192.77
(3.61)**
638.62
(4.91)**
E
2
1948.41
(2.02)*
1036.97
(4.00)**
-192.46
(-1.04)
-91.75
(-0.78)
-457.48
(-1.95)
-271.17
-1.85)
-64. 13
(-0.37)
-27.50
(-0.29)
-616.15
(2.05)*
1140.55
(7.46)**
1
A
1
A2
S1
o
o
1
.
101.16
137.21
196.67
(0.29)
(1.48)
03
47.35
(0.13)
-291.04
(_2.71)**
Y
0. 7293
(14. 07)**
0. 7789
(58. 05)**
2
R1Y1
E1Y1
E2Y1
A1Y1
A2Y1
-0. 2033
-0. 0234
(-7.22)**
(-2.01)*
-0. 1340
(_2.67)**
(-3.56)**
-0. 1708
-0. 0609
(-1.41)
-0. 1174
(_3.97)**
0.0372
(0.99)
-0.0383
(-2.27)*
0.0602
-0.0759
(-3.15)**
(1.18)
Continued
Table 13--Continued.
Regressors
SY
1
1
01Y1
02Y1
03Y1
R
2
a
Farm
Nonf arm
families
families
-0. 1806
(_4.93)**
-0.0860
(-6.06)**
-0.2076
(-3.92)**
(-7.01)**
-0. 1536
0.0033
-0. 0242
(0.04)
(-1.33)
-0. 0122
0. 0632
(-0.17)
(3.70)**
0.49
0.67
1611
8607
The figures in the parentheses are the t-values.
** Significant at the 1 % level
* Significant at the 5% level
Nonfarm families:
The negative values for regional and educational variables,
suggest that a change to the regions or educational level specified by
the variables result in a tendency for the marginal propensity to consume to decrease. The coefficient of family size is negative and significant. This is the same as that of the farm families equation.
The negative coefficient of the age variables indicates that as the
family head increases in age, the marginal propensity to consume
decreases. The coefficients for all occupational variables are
significant.
49
To test the hypothesis that there are no significant differences in
the marginal propensity to consume of farm and nonfarm families if
they have homogenous characteristics with respect to family size,
the age of family head, the educational level of family head, the
occupation status of family head, and region, one would need to
disaggregate the farm and the nonfarm family samples into sub-
groups with common characteristics of these particular variables.
The procedure used in this study is based on the socioeconomic
variables specified in equation (4. 38). All families which have the
same values for the variables specified are grouped together. As
a result, each of the groups is homogeneous with regard to a set of
socioeconomic variables. Information on the groups with at least
15 observations in the farm and the nonfarm samples is presented
in Table 14.
The distribution of families by selected characteristics of
the families in the farm and nonfarm samples is shown in Table
15.
Eighty-three percent of families in the farm sample lies
in the 17 selected family types; but it constitutes only 46 percent
of the families from the nonfarm sample. The t-test is employed
to test the null hypothesis that this percentage is the same in the
For distribution of families in all groups, see Appendix B.
Table 14. Socioeconomic characteristics of the family in each group.
Group
No.
Region
Socioeconomic characteristics of family within each group
Education level of
Family size
Age of family
family head
head
Occupation of family
head
Less than 12 years
More than 12 years
Less than 12 years
More than 12 years
Less than 12 years
2. 0-2.9 persons
2.0-2.9 persons
2. 0-2.9 persons
4
5
Northcentral and Western
Northcentral and Western
Northcentral and Western
Northcentral and Western
Northcentral and Western
More than 3. 0 persons
More than 3.0 persons
Less than 54 years
55-64 years
More than 65 years
Less than 54 years
Less than 54 years
6
Northcentral and Western
Less than 12 years
Less than 3.0 persons
Less than 54 years
7
8
9
10
11
Less than 12 years
13-16 years
Less than 12 years
Less than 12 years
Less than 12 years
Less than 12 years
Less than 12 years
More than 3. 0 persons
More than 3.0 persons
2. 0-2.9 persons
13
Northcentral and Western
Northcentral and Western
Northeast and Southern
Northeast and Southern
Northeast and Southern
Northeast and Southern
Northeast and Southern
More than 3. 0 persons
Less than 3. 0 persons
55-64 years
Less than 54 years
Less than 54 years
55-64 years
More than 65 years
Less than 54 years
Less than 54 years
14
Northeast and Southern
Less than 12 years
More than 3. 0 persons
Less than 54 years
15
Northeast and Southern
Less than 12 years
More than 3. 0 persons
Less than 54 years
Self-employed
Self-employed
Self-employed
Self-employed
Self-employed
Self-employed
Professional and managers
and retired
Clerical and sales and
unskilled workers
Skilled and semi-skilled
16
17
Northeast and Southern
Northeast and Southern
Less than 12 years
13-16 years
More than 3.0 persons
More than 3.0 persons
55-64 years
Less than 54 years
Self-employed
Self-employed
1
2
3
12
Self-employed
Self-employed
Self-employed
Self-employed
Clerical and sales and
unskilled workers
Skilled and semi-skilled
workers
2.0-2.9 persons
2.0-2.9 persons
workers
01
0
51
Table 15. Distribution of families in the farm and the nonfarrn samples by characteristics of
the family.
Group a/
No.
Farm families
Observed
1
50
2
69
41
318
51
3
4
5
6
7
Expected
44
64
8
28
51
79
9
10
11
50
244
20
12
13
14
15
16
77
78
69
17
17
Nonfarm families
Observed
Expected
(21)
(25)
(14)
33
(119)
(124)
(256)
148
446
961
(347)
(363)
(749)
(22)
21
(63)
17)
40
27
29
25
(
30
15
(20)
(28)
(19)
(101)
61)
(185)
(290)
(297)
(180)
(543)
(848)
(69)
24
37
14)
(
No. of families in
17 selected groups
r1 = 1, 340
r2 = 3, 922
Total families in the sample
n1
1,611
= 8, 607
Percent in 17 selected groups
P
r
1
n
P2 =-n
100=83
= 100 - P1 = 17
s12
2
1
Farm families: (Obs. - Exp.) /Ecp. = 2010. 66
Nonf arm families:
(Ohs. - Exp. )2/Exp. 529.05
2
x
2
01
= 32.00, d. f. = 16
For description of the group, see Table 14.
S22 = P292/n2 = 0.289
= 48.68**, d.f. = 10216
n1
x = 2010.66 + 529.05 = 2539. 71**
100= 46
Q2 = 100 - P2 =
P191/n1 = 0.876
51)
(58)
(80)
(56)
1,060
(24)
(
(
154
221
651
(
(62)
(74)
(42)
40)
52
two samples. The calculated t-value is 48. 68. This is significant
at the one percent level with the degrees of freedom being 10216.
Thus, the null hypothesis is rejected. This implies that the distribution of families in the farm sample by family types is different
from that of the nonfarm sample. It can be easily observed from
the table that most of the farm families have the socioeconomic
characteristics of groups 4 and 12, while most of the nonfarm
families have Number 6 and 15 group& characteristics.
One may also want to know whether the distribution of farm
families within the 17 selected family types is different from that
of the nonfarm sample. To investigate this question, we set up the
null nypothesis: both farm and nonfarm families in the 17 selected
groups have the same probability distribution. For the one percent
level of significance, the critical X2 value is 32. 00 which is less
than the calculated X2 value of 2539. 71. Thus, we reject the null
hypothesis and conclude that farm and nonfarm families in the 17
selected family types have significantly different distributions.
The consumption function for each group is presented in
Table 16. The t-test is used to test the quality of the marginal
propensity to consume between farm ar*d nonfarm families. The
results of testing is presented in the last column of the table.
One noticeable result is that most farm and nonfarm families
with similar characteristics have no significant differences in their
Table 16. Consumption functions and related statistics for selected homogeneous groups of farm and nonfarm families in the U. S.,
1961.
Group
No.
1
2
3
4
Farm families
1228.40+0.1378 Y
(2.57)* (1.58) 1
C1= 1035.94+0. 1750 Y1
(2.02)* (1.84)
C1 = 770.92 + 0.1980 Y1
(1.45)
(1.96)*
C1 =
C
1
S
6
7
= V292. 53 +
(2.90)**
0.3184 Y
(4.02)**1
C1=1956.03+0.5138Y1
(3.97)** (5.56)**
C1= 1908.69+0.5260Y1
(5.81)** (8.92)**
C1= 1100.07+0.3556Y1
(2.28)* (4.06)**
8
C1=2485.30+0.1844Y1
9
C1 = 205.64 + 0.3411
(4.48)**
(1.96)*
(.45)
10
11
12
(4.12)**
C1= 13. 19+ ).3783Y1
(0.03)
(4.17)**
C1= -251.83+0.4013Y
(4.13)**1
(-.49)
C1=269.78+0.5217Y1
(.64)
(7.04)**
Nonfarm families
C1 = 2351.81 + 0.5159 Y1
(10.97)** (16.33)**
C1=2260.06-f-0.4776Y1
(9.23)** (13.34)**
C1 = 2080.64 + 0.4400 Y1
(8.01)** (11.08)**
C1 = 2379.30 + 0.6019 Y1
(12.23)** (21.34)**
C1 = 947.72 + 0.8187
(5.95)** (33.15)**
C1 = 1238.76 + 0.7555
(10.52)** (42.44)**
C1 = 2287.55 + 0. 5636 y
(I0.12)** (17.13}**
C1= 3017.93+0.5410Y1
(12.97)** (16.39)**
C1 = 2250.65 + 0. 5393 Y1
(11.29)** (18.34)**
= 2158.90 + 0.5010
(9.32)** (14.78)**
C1= 1979.48+0.4634Y1
(8.00)** (12.20)**
C1 = 2278.14+ 0.6253 Y1
(12.95)** (19.42)**
Differences of MPC1 between farm
and nonfarm families
0.3781
(4.06).4°K
0.3026
(2.97)**
0. 2420
(2.22)*
0.2835
(3.36)**
0.3049
(3.18)**
0.2295
(3.71)**
0.2080
(2.22)*
0.3566
(3.57)*
0. 1982
(1.90)
0. 1227
(1.26)
0.0621
(.59)
0.1036
(1.32)
Continued
Ui
Table 16--Continued.
Group
No.
13
Farm families
C1=1023.14+0.7326Y
(1.83)
14
C1=933.28+0.7171Y1
15
C=
16
C1 =
(1.99)* (8.15)**
885.93 + 0.7293
(3.02)** (14.08)**
77.32 + 0.5589 Y1
(.17)
17
(7.65)**1
(6.73)**
C1= 1462.55+O.3877Y1
(2.73)** (4.33)**
Nonfarm families
C1= 1334.26-i-0.7547Y1
(8.39)** (33.39)**
C1=846.56+0.8421Y1
(6.21)** (38.81)**
C1= 1137.60+0.7789Y1
(13.08)** (58.13)**
C1 = 2186.39 + 0.5870 Y
(10.33)** (19.12)*4
C
1
=2917.77+0.5644Y
(13.33)** (18.27)*4
Differences of MPC1 between farm
and nonfarm families
0.0221
(0.22)
0.1250
(1.38)
0.0496
(0.92)
0.0281
(0.32)
0.1767
(1.06)
The figures in the parentheses are the t-values.
** Significant at the 1% level
Significant at the 5% level
*
Ui
55
marginal propensities to consume. This supports the first hypoth-
esis of this study- -there are no significant differences in the marginal propensity to consume for farm and nonfarm families if they
have honogeneous characteristics. This also provides evidence for
the prediction of the theory of consumer choice that the marginal
propensity to consume is a function of the utility function. If the
families are honogeneous with respect to their preferences, their
marginal propensity to consume is expected to be the same.
It is also observed that homogeneous farm and nonfarm
families in the Northcentral and Western regions have significant
differences in their marginal propensities to consume. These are
the exception to the above assertion, Puterbaugh [24] studied differences between prices paid by farmers and nonfarmers for the
goods and services they consume. The results showed that there
were no differences between prices paid by farm and nonfarm groups
for nonfood items in all regions. However, food prices paid by
farm groups were different from nonfarm groups in the Northcentral
and Western regions. Thus, in these two regions, observed differences in the marginal propensities to consume between farm and
non-
farm families may be due to the existence of price differentials on
food items between the two groups. This suggests that the assumption of constant prices among consumers in cross sectional data may
be invalid. The price variable plays an important role in the con-
sumption decision because it affects the budget constraint and
56
real income. According to the theoretical framework presented
in Chapter II, it affects the marginal propensity to consume of
consuming units.
Marginal Propensity to Consume Out of
Total Disposable Income
The marginal propensities to consume based on disposable
money income for farm and nonfarm families have been discussed
in the preceding section. However, farm families generally have
a larger amount of nonmoney income as compared to that of non-
farm families. Thus, an attempt is also made to estimate the
marginal propensity to consume based on total disposable income
which is composed of both disposable money income and disposable
nonmoney income. The marginal propensity to consume out of total
disposable income is defined as the current response of consumption
to an additional dollar of total disposable income during the survey
year. The empirical results are discussed in this section.
In the urbanization equation, all coefficients are significantly
different from zero at the one percent level. The R2 for the equation is 0.67. This is slightly higher than the R2 of equation (4. 1)
in which disposable money income is used as an explanatory vari-
able and the dependent variable is money expenditure.
In equations (4.41), (4.42), and (4.43), the coefficient of total
57
disposable income (Y2) is the marginal propensity to consume
out of total disposable income. For farm families, the marginal
propensity to consume out of total disposable income is 0.5511
which is higher than their marginal propensity to consume out of
disposable money income (0. 4649). However, for rural nonfarm
and urban families, the marginal propensities to consume out of
total disposable income are 0. 7608 and 0. 7800, respectively.
These estimates are not much different from the marginal propensities to consume out of disposable money income indicated in
equations (4. 3) and (4. 4).
One noticeable feature observed from the comparisons is that
the extent of differences between the two types of the marginal
propensities to consume for farm families is greater than that of
nonfarm families. This may reflect the fact that farm families
have a larger portion of nonmoney income than that of rural non-
farm and urban families.
The differences in the marginal propensities to consume out
of total income for farm, rural nonfarm, and urban families and
their t-values are given in Table 18. The marginal propensity to
consume out of total disposable income for rural nonfarm families
is not significantly different from that of urban families. How-
ever, the estimates for these two types of families are significantly
different from that of farm families. These findings are the same
Table 17. Regression coefficients and related statistics of urbanization
equation and three urbanization families in the U. S., 1961.
Constant term
Equation
U1
U2
U1Y2
U2Y2
0. 2097
0. 2289
(11.09)**
(15.54)**
R2
Urbanization:
(4.40) C 2
1584.94
(19.93)**
-679.35
(_5.85)**
-561.88
(-6.03)**
0.5511
(41.92)**
Rural nonfarm families:
905.58
(4.41) C2
(5. 03 )**
0. 7608
(55.94)*K
1023.05
(6. 15)**
(116.42)**
1584.94
(19.93)**
(41.92 )**
Urban families:
(4.42) C2
0. 7800
Rural farm families:
(4.43) C2
where
0.5511
C2 = total expenditures for current consumption
Y2 = total disposable income
U1 =
1,
if the rural nonfarm families;
0, otherwise
U2 = 1, if the urban families; = 0, otherwise
** Significant at the 1% level.
0.67
59
Table 18. Differences in the marginal propensities to consume
out of total disposable income and their t-values for
farm, rural nonfarm and urban families in the U.S.,
1961.
Rural nonfarm
Rural nonfarm
Urban
Farm
0
(0)
Urban
Farm
-0.0192
(-1.27)
0. 2097
(11. 09)**
0
(0)
0. 2289
(15. 54)**
0
(0)
as those estimates based on disposable money income. Therefore,
it also supports the conclusions of previous studies that farm
families respond differently to consumption as income changes
than do nonfarm families when other socioeconomic factors are
not taken into consideration.
After substitution of total income (Y2) for money income (Y1)
and total expenditure (C2) for money expenditure (C1), equations
(4. 38) and (4. 39) are fitted to the data by multiple least squares
regression. The multiple regression coefficients and related
statistics estimated for farm and nonfarm families are presented
in Table 19. In general, the regression coefficients estimated
are about the same as those estimated based on money expenditure
and disposable money income (Table 13). The R2 for farm fam-
ilies equation is 0. 56 which is greater than the R2 estimated in
Table 19. Multiple regression coefficients and related statistics in farm and nonfarm families,
U.S., 1961.
Regressors
Constant
R1
E1
Farm
Nonfarm
families
families
614. 17
2670.53
(2.33)*
-149.25
890.64
(10.00)**
112.58
(1.38)
273.70
(2.07)*
765.87
(2.79)**
-111.05
(-0.68)
(-0.92)
-493.05
(-1.85)
-397.54
(-2.65)**
28.11
(0.14)
(1.05)
1040.02
(3.40)**
490.60
(0.93)
149.71
(0.39)
1299.07
(8.22)**
373.51
(2.74)**
-217.27
(.-1.97)*
(2.05)*
894.94
(5.33)**
-428.62
(-1.32)
E2
A1
A2
S1
03
R1Y2
E1Y2
E2Y2
A1Y2
A2Y2
0. 8076
0. 8270
(17.29)**
-0. 1651
(-5.87)**
(63.01)**
01Y2
-0. 0265
(-2.30)*
0. 1268
-0. 0127
(3.00)**
(-0.77)
-0. 2412
-0. 796
(-1.86)
(_4.97)**
(-2.67)**
-0.0392
(_2.36)*
-0.0432
(-1.85)
-0.981
(-7.06)**
-0. 2397
(-5. 10)**
(_7.97)**
0.0251
(0.64)
0. 0681
(1.35)
S1Y2
100.89
-0. 1909
-0. 1733
Continued
61
Table 19- -Continued.
Regressors
02Y2
Farm
Nonf arm
families
families
-0.0593
-0.0516
(-2.90)**
(-0.74)
03Y3
-0. 0166
n
(-0.25)
0. 0529
(3. 17)**
0.56
0.70
1611
8607
The figures in the parentheses are the t-values.
** Significant at the 1% level
* Significant at the 5°/s level
Table 13 in the previous section. This implies that the farm fam-
ilies equation has a better fit by using total income and total ex-
penditure variables. However, the R2 in nonfarm families equation
is 0. 70. This is not much different from that estimated based on
disposable money income.
Consumption functions and related statistics for selected
homogenous groups of farm and nonfarm families are presented
in Table 20. Differences in the marginal propensities to consume
between farm and nonlarm families in each group were subjected
to a t-test. Eleven out of 17 groups of farm and nonfarm families
with homogenous characteristics had no significant differences in
their marginal propensities to consume out of total disposable
income. The groups with significant differences in the marginal
Table 20. Consumption functions and related statistics for selected
homogeneous groups of farm and nonfarm families in the U.
Group
No.
1
2
3
4
5
6
7
8
9
10
Farm families
C2=2577.24+0.2119Y2
(5.14)** (2.60)**
C =
2
=
2403. 18 + 0.5291 Y
(10.89)** (l7.Ol)J
2292. 12 + 0.4899 Y
(9.11)** (13.88)*
C2=2427.98-f-0.2370Y2
(4.42)** (2.62)**
C2
C2=2084.19+0.2800Y2
C2= 2005.63+ 0.4859 Y
3.66)** (2.92)**
C2 = 2549.12 + 0. 4028
(5.54)** (5.59)**
=
C2
=
C2=1658.82+0.62$9Y
C2
C2= 1509. 11+0.6425Y
C
(3.20)** (7.20)**2
(4.39)** (11.79)*
C2=2399.87+0.4279Y2
C
(4.71)** (S.21)**
C2 = 2120.51 + 0. 5296
(3.76)** (6.34)**
C
C2= 1682.30+0.3770Y2
C
(3.55)** (4.92)**
C2= 1533.04+0.4021Y2
(2.93)**
11
Nonfarm families
(4.67)**
C2= 1189.25+0.4451Y2
(2.19)*
(4.85)**
=
2
(7.51)** (12.46)*
2302.29 + 0.6272 Y
(11.58)** (22.48)*
785.95 + 0. 8534 Y
(4.81)** (35.27)*
1003.22+0.8005Y
(:830)**
(46.01)*
2
=2191.24+0.5880Y
2
=2575.99+0.6145Y2
(9.42)** (18.15)*
(10.79)** (18.97)**
=
2
2290.60 + 0. 5556
Y2
(11.17)** (19.16)**
C
2
=2179.54+0.5164Y2
C =
2
(9.16)** (15.46)**
1893.05+0.5124Y
(7.45)** (13.77)*
S., 1961.
Differences of MPC2 between farm and
nonfarm families
0.3172
(3.76)**
0.2529
(2.69)**
0.2059
(2.06)*
0.2244
(3.01)**
0. 2275
(2.62)**
0. 1580
(2.86)**
0. 1601
(1.88)
0.0849
(0.98)
0. 1786
(1.89)
0.1143
(1.28)
0.0673
(0.70)
Continued
Table 20- -Continued.
Group
No.
12
Farm families
C2 = 1654. 18 + 0.5679
Nonfarm families
C2 = 2189.71 + 0.6537 Y
0.0858
(12.07)** (25.74)*
1264. 15 + 0.7754 Y
(7.77)** (35.09)*
673.37 + 0.8799
(4. 75)** (41.31)
890.64 + 0. 8270 Y
(1.25)
(3.86)** (8.58)**
13
C2= 1104.77+0.7483Y
(8.04)**2
(1.82)
C2 =
14
C2=763.88+0.7910Y
C2 =
15
(1.56)
(9.62)*
C2= 614.17+ 0.8076Y
(2.05)* (17.29)*
C2 =
16
C2 = 1504.93 + 0. 5930
C2 = 2073.66 + 0.6145 Y2
17
C2= 1225.57+ 0.6947 Y
(2.28)* (8.84)**2
(3. 13)**
(7.69)**
Differences of MPC2 between farm and
nonfarm families
(10.00)** (63.13)*
(954)** (20.21)**
C2 = 2463.41 + 0.6410 Y
(12.39)** (21.16)*
0.0271
(0.29)
0.0889
(1,09)
0.0194
(0.42)
0.0215
(0,26)
-0. 0537
(-0.66)
The figures in the parentheses are the t-values.
** Significant at the 1% level
* Significant at the 5% level
C'
64
propensities to consume between farm and nonfarm families are
located in the Western and Northcentral regions. This is parallel
to the preceding findings.
In summary, the estimates of the marginal propensities to
consume out of total disposable income for rural nonfarm and urban
families are comparable to their marginal propensities to consume
out of disposable money income. However, for farm families the
marginal propensity to consume is higher than the estimate based
on disposable money income. This may be due to the fact that farm
families have a significant amount of disposable nonmoney income
in addition to disposable money income. The findings indicate that
most farm and nonfarm families have no significant differences in
their marginal propensities to consume out of total disposable income if both families are homogenous in socioeconomic character-
istics.
65
V CONCLUSIONS AND IMPLICATIONS
This study demonstrates some significant relationships between socioeconomic characteristics of the family and the marginal
propensity to consume. As a first step a theoretical framework was
presented which indicated the relationship between the theory of
consumer behavior and the marginal propensity to consume. From
this framework, a set of hypotheses was formulated for explaining
observed differences in the marginal propensities to consume between
farm and nonfarm families. Data from the Bureau of Labor Statistics and the U. S.. Department of Agriculture--Survey of Consumer
Expenditure Conducted in 1960-61 were then used in conducting
consumption function analysis by the statistical procedure of least
squares regression. The results were presented in the preceding
chapter. This chapter is to draw some conclusions with respect to
(1) validity of hypotheses, (2) comparisons with previous studies,
and (3) research methodology. It also presents policy implications
and suggestions for future research.
Validity of the Hypotheses
Hypothesis #1: Farm and nonfarm consuming units with
homogenous socioeconomic characteristics have the same marginal
propensity to consume.
The empirical results (Tables 16 and 20) indicate that there
are no significant differences in the marginal propensities to
consume between farm and nonfarm families with identical socio-
economic characteristics for the majority of the 17 tested groups.
However, those groups in the Northcentral and Western regions
were the exception. In these two regions, farm families have a
lower marginal propensity to consume than that of nonfarm families.
This appears to be due to the fact that prices paid by farm and non-
farm families are different. This suggests that the assumption of
constant prices among consumers in cross sectional data may be
invalid. The price variable plays an important role in the con-
sumption patterns because it affects the budget constraint and real
income.
Hypothesis #2: Distribution of consuming units by socio-
economic characteristics in the farm sample is different from that
of consuming units in the nonfarm sample.
The distribution of families based on socioeconomic charac-
teristics for farm and nonfarm samples was shown in Table 15.
A t-test indicated that this hypothesis is accepted. The difference
in distribution of families in both samples can also be observed
from Table 21. There was a higher percentage of families headed
by persons with low educational levels, self-employed, and with
more children in the farm sample than that in the nonfarm sample.
67
Table 21. Distribution of families in the farm and the nonfarm samples based on socioeconomic
characteristics.
Socioeconomic
characteristics
Urban
N
N
Rural nonf arm
%
Total nonf arm
N
%
Farm
N
°/
Regions:
Northcentral
and Western
Northeast and
Southern
3277
48
709
40
3986
3576
52
1045
60
4621
54
854
53
Educational level:
Less than 12 years
More than 13 years
5276
1577
77
23
1498
85
6774
15
1833
79
21
1485
126
92
256
5161
984
708
75
14
1273
72
6434
241
1225
75
14
240
948
11
1064
362
185
66
22
11
14
14
2263
33
555
32
2818
33
451
28
4590
67
1199
68
5789
67
1160
72
468
6
195
11
663
8
1184
74
1961
29
444
25
2405
28
74
4
2582
38
679
39
3261
38
178
11
1842
27
436
25
2278
26
175
11
Ages:
Less than 54 years
55 to 64 yrs. old
Over 65 yrs. old
Family size:
2.0 to 2.9 persons
More than 3.0
persons
Occupation:
Self-employed
46
757
47
8
12
Professional,
manager and
retired
Skilled and
semi-skilled
workers
Clerical, sale
and unskilled
workers
Total families in
sample
6853
1734
8607
1611
However, there was almost no difference in the distributions of
families by socioeconomic characteristics between urban and rural
nonfarm samples. The empirical results (Table 2) indicated that
the marginal propensity to consume for farm families was significantly different from that of nonfarm families; but there were no
significant differences in the marginal propensities to consume
between urban and rural nonfarm families. Thus, observed differ-
ences in the marginal propensities to consume between farm and
nonfarm families could be due to the difference in distribution of
family types in the two samples.
Comparison with Previous Studies
(1) Current study vs. Friedman's study
The results of the estimates from the urbanization equations
(4. 1) and (4. 40) of the current study are compared with those in
the Friedman study (Table 22). However, it should be noted that
there are some difficulties in making a comparison between these
two studies because of differences in methodology. First, the
logarithmic form of the consumption function for each urbanization
group was used in the Friedman study, while the linear form was
used in the present study. Second, Friedman used gross income
as regressors while the present study used disposable money income and total disposable income. Third, individual family
Table 22. Comparison of current study and Friedman's study: the marginal propensities
to consume of farm, rural nonfarm, and urban families in the U. S.
Family
Current study
(1961)
groups
Friedman's study
(1935-36)
MPC1'
MPC2'
MPC3"
Farm
0.4649
0.5511
0.50
Urban
0. 7393
0. 7800
Rural nonfarm
0. 7255
0. 7608
0. 73
(1941)
MPC3"
0.48
d/
0. 79
J
MPC1 = marginal propensity to consume out of money disposable income.
MPC2
marginal propensity to consume out of total disposable income.
MPC3 = marginal propensity to consume out of total gross income.
Nonfarm families.
Table 23. Comparison of current study and Lee and Phillips' study: the income elasticities of
farm, rural nonf arm, and urban families in the U. S.
Family
Current study
groups
Lee and Phillips
TI
2
1
2
Farm
0.5599
0.6458
0.519
Urban
0. 7951
0. 8389
0. 744
Rural nonfarm
0. 7894
0. 8278
0. 742
income elasticity on total money expenditure.
dC1
= MPC1
TI
b/
TI
2
TI2
1
= income elasticity on total expenditure.
MPC2
APC2
70
observations were used in the regression analysis of the current
study while group mean data were used in the Friedman study.
The marginal propensities to consume estimated in the current study, without taking other socioeconomic variables into
account, support Friedman's findings that farm and nonfarm fam-
ilies have different marginal propensities to consume. However,
when some socioeconomic characteristics of the family were taken
into consideration, the current study verified that the farm and nonfarm families do not necessarily have different marginal propensi-
ties to consume. Furthermore, this study indicates that observed
differences in the marginal propensities to consume between farm
and nonfarm families as reported in the Friedman study could be
due to differences in the distribution of families by socio-economic
characteristics in the two samples.
(2) Current study vs. Lee and Phillips' study
In a study of differences in consumption patterns of farm and
nonfarm households, Lee and Phillips used group mean average
data from the Survey of Consumer Expenditure conducted by U. S.
Department of Agriculture and Bureau of Labor Statistics in
1960-61.
The two stage least squares method was employed to
estimate the parameters of their model. In contrast, in the current
study all observations and the ordinary least squares method were
used. In addition, the current study has different model
71
specifications than that of Lee and Phillips.
To compare income elasticities estimated by Lee and Phillips,
the marginal propensities to consume estimated from the urbanization equations (4. 1) and (4.40) have been converted into income elas-
ticities. The empirical results of the current study are different
from those of the Lee and Phillips study (Table 23). This could be
attributed to differences in the model specification and the statistical
procedures used in the two studies.
Methodological Conclusions
(1) Socioeconomic variables and the dummy variable technique
The conventional treatment of other socioeconomic variables
(i. e., other than income and prices) has been to combine them in a
general variable described as variations in taste and preferences.
This group of variables has often been converted under the econo-
mists' assumption of ceteris paribus (other variables remain Un-
changed), or, statistically, this group of variables has been included in a residual term because these variables are not continuous and difficult to quantify. However, this study adopts the dum-
my variables technique which affords a means of quantifying other-
wise nonquantifiable variables and thus of measuring their net effects.
The essence of the technique involves assigning a dummy variable
to all categories of a characteristic except one. Because the
72
variable takes a value of one if the observation belongs to that
category and zero, otherwise, it is called a dummy variable.
Through the use of dummy variables in regression analysis
the effects of certain socioeconomic variables were estimated. The
results show these variables are important in explaining variation
in the marginal propensity to consume.
(2) The use of group mean data and individual observations
Previous studies concerning differences in the consumption
patterns of farm and nonfarm families in the United States have used
group mean data in their analysis; however, the current study used
individual observations from the household survey. There are
several disadvantages in using group mean data as compared to
individual observations.
Because grouping sharply reduces the number of degrees of
freedom, tests of significance become somewhat less powerful.
A larger variance of the regression coefficients will result because
of the decrease in the degrees of freedom. The variance of regression coefficient (p.) is equal to
2
a-
times its corresponding main
inverse matrix element. But the error variance estimate (a- 2) is
equal to the residual sum of squares (e2) divided by the degrees
of freedom. Thus
2
a-
is biased upwards which in turn causes a
larger variance for the regression coefficient (var
73
The coefficient of determination (R2) is vastly overstated when
a consumption function is estimated by the use of group mean data
rather than individual observations. This is so because grouping
always reduces sharply the scatter of observations. Therefore, it
increases the correlation between two related variables. There
exists a tendency that the fewer the observations used in an analysis
the larger the overestimation of R2
Of course, the use of group mean data in the study costs much
less as compared to using individual observations. However, if the
precision and the efficiency of estimations are of primary importance, the
information
from all individual observations should be
used.
Policy Lmplications
The empirical analysis of Chapter IV indicates that the marginal propensity
to
consume out of disposable money income is
0. 4649 for farm families; 0. 7255 for rural
nonfarm
families; and
0. 7393 for urban families. For farm families the marginal pro-
pensity to consume out of total disposable income is 0.5511; for
rural
nonfarm
families it is 0. 7608; for urban families it is 0. 7800.
This evidence shows that there are
farm and
nonfarm
significant
differences between
families in their consumption responses to
changes in their income. It also suggests that each dollar shifted
74
from farm to nonfarm families will increase consumption by about
22 cents without the multiplier effect. Therefore, there appears
to exist a redistribution effect from governmental taxation and expenditure upon aggregate consumption.
However, when some socioeconomic variables are taken into
consideration, the empirical analysis shows that there is no significant difference in the marginal propensity to consume for farm and
nonfarm families with similar characteristics except in the regions
of the West and Northcentral. This suggests that to carry out a
redistribution income policy, farm and nonfarm sectors are not a
proper criteria at the national level. The differences in the marginal propensity to consume for farm and nonfarm families exist in
the Northcentral and Western regions. This implies that the re-
distributional effect between two sectors in these two regions will
be effective but not in the other regions.
In summary, the results from this study seem to caution
against placing too much reliance on the theory that income re-
distribution between farm and nonfarm sectors will greatly increase
the volume of consumption in the economy. The study also shows
that governmental actions for redistribution of income between farm
and nonfarm sectors can be effective in certain regions only. This
is necessary to emphasize in regional research and regional
economic planning.
75
Implication for Future Research
(1) A limitation of inferring econometric relationships from
observations gathered in a single survey is that prices are assumed
to be constant over the survey period and to be the same for all consumer units. In fact, there are existing geographical price differ-
entials. It is known that the price variables play an important role
in consumption patterns. This can be investigated only through the
analysis of time series data, Thus, the integration of crosssectional and time-series data in explaining the differences in the
marginal propensity to consume of farm and nonfarm families is
another area where significant advances can be expected from fur-
ther research.
(2) This study is a cross-section analysis at one point in time,
early 1961, and may not be applicable to other points in time. The
Bureau of Labor Statistics and the U. S. Department of Agriculture
are undertaking a consumer expenditure survey for 1972 by using
the same sampling method and definitions are used in the 1961 survey.
Therefore, it may be worthwhile to conduct a similar study
based on the 1972 survey. It would provide estimates at another
point in time for comparison and permit one to determine the extent
of changes over time.
ii1
BIBLIOGRAPHY
1.
Batie, S. S. and Hammonds, T. M. MuRicollinearity
Problems Encountered When Testing for Regression Equation Equality, Corvallis, Oregon, Department of Agricultural Economics, Oregon State University, 1972. (unpub1 is he d)
2.
Brown, W. G. Effect of Omitting Relevant Variables in
Economic Research. Corvallis, Oregon. Oregon Agricultural Experiment Station Technical Paper No. 2723,
Research conducted under Project 845. 1968.
3.
Brown, W. G. and Nawas, F. Improving the Estimation
and Specification of Outdoor Recreation Demand Functions,
Corvallis, Oregon, Oregon Agricultural Experiment Station
Project 850, 1971. (unpublished)
4.
Burke, M. C. Food Expenditures by Upper Income Families,
University of Minnesota, Agricultural Experiment Station in
cooperation with the U. S. Department of Agriculture, Technical Bulletin 269, 1969. p. 121 -123.
5.
Consumption Economics: A Multidisciplinary Approach. New York, John Wiley and Sons,
Inc. 1968. p. 81-83.
6.
Chow, G. C. Tests of Equality Between Sets of Coefficients
in Two Linear Regressions. Econometrica 28:591-601,
July 1960.
7.
Christ, C. F. Econometric Models and Methods. New York,
John Wiley, 1966. p. 102-106.
8.
Cramer, J. S. Efficient Grouping, Regression and Correlation in Engel Curve Analysis. Journal of American Statistical Association 59:233-250. March 1964.
9.
Crockett, J. and Friend, I. A Complete Set of Consumer
Demand Relationships. In: Consumption and Savings, ed.
Friend, I, and Jones R. Wharton School of Finance and
Commerce, University of Pennsylvania, 1961. Vol. 1,
p. 1-92.
L
77
10.
Freund, R. 3. Some Observations on Regressions with
Grouped Data. The American Statistician 25:29-30.
June 1971.
11.
Friedman, M. A Theory of Consumption Function.
Princeton, Princeton University Press, 1957. p. 58-75.
12.
Gujarati, D. Use of Dummy Variables in Testing for
Equality Between Sets of Coefficients in Linear Regression:
A Generalization. The American Statistician 24:18-22.
December 1970.
13.
Henderson, J. M. and Quandt, R. E. Microeconomic
Theory. New York, McGraw-Hill, 1971. p. 14-15.
14.
Johnston, J. Econometric Methods. New York, McGrawHill, 1963. p. 221-228.
15.
Karnenta, J. Elements of Econometrics, The Macmillan
Company, New York, 1971. p. 419-423.
16.
Klein, L. R. The British Propensity to Save. Journal of
the Royal Statistical Society, Ser. A (General), Part 1.
121:60-95. 1958.
17.
An Introduction to Econometrics,
Evanston, Ill., Row, 1953. p. 52-60.
18.
Laumas, P. 5. A Test of the Permanent Income Hypothesis.
Journal of Political Economy 77(5):859-861. September!
October 1969.
19.
Lee, F. Y. and Phillips, K. E. Differences in Consumption
Patterns of Farm and Nonfarm Households in the United
States, American Journal of Agricultural Economics 53:
573-582. November 1971.
20.
Liviatan, N. Errors in Variables and Engle Curve Analysis.
Econometrica 29:336-362. June 1961.
21.
Malinvaud, E. Statistical Methods of Econometrics.
Chicago, Rand McNally and Company, 1966. p. 129-135.
22.
Mendershausen, H. Differences in Family Savings
Between Cities of Different Size and Location, White
and Negroes. Review of Economic Statistics 22:122-137.
August 1940.
23.
Ostile, B. Statistics in Research. 2nd Edition. Ames,
The Iowa State University Press, 1963. p. 201-205.
Iowa.
24.
Puterbaugh, H. L. Purchasing Power of Urgan, Rural
Nonfarm, and Rural Farm Income, 1955. Agricultural
Economic Research 53:89-94. July 1961.
25.
Suits, D. The Use of Dummy Variables in Regression
Equations. Journal of the American Statistical Association
52:548-551.
26.
U. S. Department of Labor, Bureau of Labor Statistics,
Consumer Expenditures and Income: Survey Guidelines,
Bulletin 1684, Washington, D.C. 1971. p. 11-16.
27.
U.S. Department of Labor, Bureau of Labor Statistics,
Survey of Consumer Expenditures, 1960-61: Descriptive
Materials on Use of CES General Purpose Tape. Washington,
D.C. February 1972.
28.
U.S. Department of Labor, Bureau of Labor Statistics, and
U. S. Department of Agriculture, Agricultural Research
Service, Survey of Consumer Expenditures and Income,
USDA-CES Reports 1-45 and BLS Reports 237:84-93.
Washington, D.C. 1965-1966.
APPENDICES
79
APPENDIX A
The Equality of Regression Coefficients of the Dummy
Variable Regression Model and the Ordinary
Regression Model 7/
The consumption function for farm and nonfarm families is:
(1)
C
a + a1D +
Y + t31DY (farm and nonfarm families
N1 + N2 = N observations)
where C represents consumption, Y represents income, and
D is a dummy variable such as D=1, if the observation belongs
to nonIarm families
0, otherwise, farm families.
Then, we would have
+ 3Y
(2)
C
(3)
C = (a+ a1) +
(farm families, N1 observations)
+
)Y
(nonfarm families, N2 observtions)
The least squares estimators of the regression coefficients of
equation (1) are exactly the same as those that would be obtained
from two separate regressions of C on Y, one estimated from the
farm families (C = a0 + b 0Y) and the other from the nonfarm families (C = a1 + b1Y).
That is
For an empirical example, see [i, p. 6].
a0 =a 0
a0 +a =a
1
+
b1
Proof: The proof is based upon a straightforward application of
the least squares formulas. For convenience it is broken into
three steps.
I.
We first obtain the sum of squared residuals from the equation
(1):
N
e
(4)
1
Setting
2
N
=
N
=
1
0
it gives:
0
N
(5)
1D) Y]2
+
2
1
8'a
E [C - (a0+ a 1D)
N
(a° + a1D) +
N
+ p1D) Y
C
=
1
1
However, if D=0, the equation (b) becomes
N
(6)
1
a+
N
pY=E C
1
If D=l, the equation (5) becomes
N2
(7)
N2
N2
E(a+a)+Z(+1)Y=E C
0
0
1
1
1
N2
e2
1
Similarly, setting
N
(8)
= 0, we have
N
(a
10
+ a D) Y +
(p
0
1
+
3 D) Y
N
2
CY
1
However, if D0, equation (8) would become
N1
(9)
N1
N1
aY+
10
10
CY
1
If D=1, then, we have:
N2
(10)
N2
E (a+a1)Y+
1
and
N1
N1
z c-p
(11)
=
0
(p+p1)Y2
1
Solving for
1
N1
EY
01
N2
CY
=
1
from equations (6) and (9), we have
N1
N1
N
N1
CY- CY
1
(12)
N1
1
(Y-Y) (C-C)
1
1
N
12
N
_j
\z
\1 N1
(Y-Y)2
1
/
Solving for (a 0 + a ) and
1
+ P1) from equations (7) and
(10) we obtain
N2
N2
C
(l3)ao +a
=
+ p1)
(1
Y
1
1
N2
1
N2 N2
N2
CY 1
(14)
1
N2
N2
Y2-1
1
II.
(Y - Y) (C
1
EY
N2
N2
CEY
1
N
2
(Y
N2
C = a + b Y (farm families N1)
N
N
a'e2
by setting
1
1
e2
0
0
we have two equations:
and
=0,
1
o
-Y)2
C)
N1
(15)
N1
N1
a
2
1
°
b° Y=
+
1
N1
N1
a °Y +
(16)
1
C
1
N1
b
1
° Y2
CY
=
1
Solving for a 0 and b from the above equations, we have
0
N1
E (Y-Y)(C-C)
(17)
b
o
=
1
-
N1
(Y-Y)
1
N1
N1
C-b
a
(18)
III
o
Y
Q.E.D.
N1
a1 + b Y
C
:
01
(nonfarm families N2)
Similarly, we can have normal equations from the equation
of nonfarm families as follows;
N
N
2
a1+
(19)
1
N
2
b1Y=
2
C
1
N
(20)
N
2
> aY+
y2=2y
N
2
b1
1
From the normal equations, we can obtain
N2
(Y-Y) (C-C)
E
= po+pl
N2
(Y-Y)2
1
N2
N2
C-b
a1
=
1
N2
Y
a0 +a
Q.E.D.
Table 24. Distribution of families in the samples by socioeconomic characteristics.
Socioeconomic a!
characteristics
Farm
families
Rural nonfarm
families
Urban
Total nonf arm
families
families
E1A1S101R1
50
13
20
33
E1A2S101
69
9
21
30
E1A3S101R1
41
4
11
15
E1A1S201R1
318
43
105
148
E1A1S203R1
41
107
339
446
E1A1S204R1
44
183
778
961
E1A2S201R1
64
6
15
21
E2A1S201R1
68
5
35
40
E1A1S101R2
51
9
18
27
E1A2S101R2
79
15
15
29
E1A3S101R2
50
9
16
25
E1A1S201R2
244
46
108
154
E1A1S202R2
20
51
180
221
E1A1S203R2
77
147
504
651
E1A1S204R2
78
280
780
1060
E1A2S201R2
69
7
17
24
E2A1S201R2
17
6
31
37
E1A1S102R2
3
6
36
42
E1A1S102R2
4
9
42
51
7
23
135
158
10
29
169
198
l3.
E1A1S103R1
EASOR
Continued
L.z
Table 24--Continued.
Socioeconomic
characteristics
Farm
families
Rural nonf arm
families
Urban
families
Total nonfarm
families
E1A1S104R1
6
26
200
226
E1A1S105R2
8
74
213
287
E1A2S102R1
2
9
47
56
E1A2S102R2
3
9
S2
61
E1A2S103R1
2
21
66
87
E1A2S103R2
9
22
92
114
E1A2S104R1
6
18
lii
129
E1A2S104R2
6
29
96
125
E1A3S104R1
1
64
165
229
E1A3S104R2
9
69
212
281
E1A3S103R1
3
12
31
43
E1A3S103R2
2
6
28
34
E1A3S104R1
2
1
18
19
E1A3S104R2
2
5
25
30
E2A1S101R1
2
0
8
8
E2A1S101R2
5
4
4
8
E2A1S104R1
2
3
59
62
E2A1S102R2
0
5
52
57
E1A1S202R1
6
28
164
192
E1A2S202R1
2
2
18
20
E1A2S202R2
2
11
26
37
E1A3S203R1
3
7
37
44
E1A2S203R2
10
10
56
66
Continued
Table 24--Continued
Socioeconomic
characteristics
Farm
families
Rural nonfarm
families
Urban
Total nonf arm
families
families
E1A2S204R1
3
14
61
75
E1A2S204R2
9
31
101
132
E1A3S204R1
24
2
1
3
E1A3S204R2
35
3
3
6
E1A3S202R1
5
6
29
35
E1A3S202R2
0
27
67
94
E1A3S203R1
1
0
1
1
E1A3S203R2
1
8
13
21
E1A3S204R1
2
0
4
4
E1A3S204R2
2
1
6
7
E2A1S202R1
0
26
252
278
E2A1S202R2
7
46
202
250
E2A1S203R1
3
19
105
124
E2A1S203R2
4
14
113
127
E2A1S204R1
3
8
101
109
E2A1S204R2
6
14
53
67
E2A2S201R1
5
2
3
5
E2A2S201R2
2
1
1
2
E2A2S202R1
0
1
9
10
E2A2S202R2
1
1
16
17
E2A2S202R1
0
0
4
4
E2A2S203R2
0
0
5
E2A2S204R1
1
1
3
4
Continued
Table 24--Continued.
Socioeconomic
characteristics
Farm
families
Rural nonfarm
families
Urban
families
Total nonfarm
families
E2A2S204R2
1
1
5
6
E2A3S204R1
1
1
0
1
E2A3S201R2
1
0
0
0
E2A3S202R1
0
0
3
3
E2A3S202R2
0
0
2
2
E2A3S204R1
0
0
1
1
E3A1S201R1
1
1
6
7
E3A1S204R2
2
1
9
10
18
105
123
3
25
77
102
E3A1S203R1
1
0
8
8
E3A1S203R2
1
3
7
10
E3A2S201R2
1
0
0
0
E3A2S202R1
0
0
4
4
E3A2S202R2
1
1
3
4
E3A3S202R1
0
0
2
2
E3A3S202R2
0
0
3
3
E3A2S101R1
0
2
0
2
E3A2S101R2
1
1
1
2
E3A2S102R1
0
1
6
7
E3A2S102R2
1
1
2
3
E3A2S103R2
0
0
1
1
E3A1S202R1
E3A1S202R2
Continued
Table 24--Continued.
Socioeconomic
characteristics
Farm
families
Rural nonfarm
families
Urban
families
Total nonlarm
families
E3A2S104R1
0
0
1
1
E3A2S104R2
0
0
1
1
E3A3S101R1
0
0
3
3
E3A3S101R2
0
0
1
1
E3A3S102R1
0
0
4
4
E3A3S102R2
1
1
5
6
E3A3S103R1
0
0
2
2
E3A3S103R2
0
0
1
1
E2A1S102R1
0
3
41
44
E2A1S103R2
1
0
28
28
E2A1S104R1
0
0
13
13
E2A1S104R2
1
2
10
12
E2A2S104R1
8
1
6
7
E2A2S101R2
4
2
3
5
E2A2S102R1
1
4
15
19
E2A2S102R2
2
6
24
30
E2A2S103R1
0
2
9
ii
E2A2S103R2
0
2
10
12
E2A2S104R1
0
0
5
5
0
0
3
3
E2A3S101R1
1
1
1
2
E2A3S101R2
2
2
1
3
E2A2S104R2
Continued
Table 24--Continued.
Socioeconomic
characteristics
Farm
families
Rural nonfarm
Urban
families
families
Total nonfarm
families
E2A3S102R1
0
6
21
27
E2A3S102R2
0
6
29
35
E2A3S103R1
0
0
1
1
E2A3S103R2
0
1
5
5
E2A3S104R1
0
0
2
2
E2A3S104R2
0
0
1
1
E3A1S101R1
0
0
3
3
E3A1S101R2
0
0
2
2
E3A1S102R1
0
0
25
25
E3A1S102R2
1
2
19
21
E3A1S103R1
0
0
3
3
E3A1S103R2
0
0
3
3
1611
1754
6853
867
Total
Notation:
E1 =
E2 =
E3 =
A1 =
A2 =
A3 =
S1
=
S2
01
02
03
04
=
=
=
=
=
=
12 years or less of education
13 to 16 years of education
17 years or more of education
54 years old or less
55-64 years old
65 years old or more
2.0-2.9 persons in the family
3.0 persons or more in the family
self-employed
professional, manager, and retired people
clerical, sales, and unskilled workers
skilled and semi-skilled workers
Northcentral and Western regions
Northeast and Southern regions
Download